Machine Learning Approaches to Categorize Land-Use Types Using Building Electricity Consumption Data: Implementation and Performance Evaluation

This research presents an innovative machine learning approach for classifying land use types using building-level electricity consumption data. The main objective of the study is to categorize buildings into predetermined land use categories by analyzing electricity consumption patterns and subscriber information. The researchers have developed the model using advanced techniques such as XGBoost. This model has been tested in three pilot districts of Turkey (Karşıyaka, Çiğli, and Tepebaşı) with diverse urban structures. The model's performance has been evaluated using standard performance metrics such as precision and recall. Notably, ensemble learning methods (CatBoost, Gradient Boosting, and Random Forest) have consistently achieved high accuracy rates. A significant finding of this study is that electricity consumption data can be effective in classifying land use even in the absence of detailed municipal zoning data. This result is particularly important for regions where current or comprehensive zoning data is not available. The results of the research demonstrate that the use of machine learning techniques in the land-use analysis can provide significant improvements in the planning of electricity distribution networks and the project development of network investments. This approach can contribute to the development of more efficient and sustainable energy distribution systems.